The 90-Second Earnings Brief
This works with any LLM — Claude, ChatGPT, Gemini. Claude handles long documents best, so I use it for 10-K analysis. ChatGPT is fine for press releases and shorter filings.
Step 1: The Filing Dump (30 seconds)
Copy the earnings press release (or paste the 10-K section you care about) into your LLM. Then use this prompt:
You are a senior equity research analyst. I am about to join an earnings call for [TICKER]. Here is their latest earnings press release / 10-K excerpt:
[PASTE DOCUMENT]
Build me a 1-page earnings brief with these sections:
1. HEADLINE: One sentence — what is the story this quarter?
2. BEATS/MISSES: Revenue, EPS, and guidance vs consensus (if available)
3. RED FLAGS: What looks concerning? What are analysts likely to ask about?
4. TAILWINDS: What is working? What accelerated?
5. KEY QUESTION: The single most important question I should ask or listen for on this call.
Be specific. Use numbers from the filing. No generic statements.
This alone saves 15-20 minutes of prep per name. But we can do better.
Step 2: The Historical Pattern (30 seconds)
After the brief generates, follow up with:
Now compare this quarter to their last 4 quarters. What is the trend on:
- Revenue growth rate (accelerating or decelerating?)
- Margin trajectory (expanding or compressing?)
- Guidance pattern (do they sandbag or guide tight?)
- Management tone on cash deployment (buybacks vs capex vs debt paydown)
Tell me if this quarter breaks any established pattern.
Pattern breaks are where the alpha is. If a company has guided conservatively for 6 quarters and suddenly guides in-line, that is a signal. The LLM catches this faster than scrolling through old transcripts.
Step 3: The Competitive Context (30 seconds)
Given what you know about [TICKER]'s results, what does this imply for these peers: [PEER1, PEER2, PEER3]?
Specifically:
- If [TICKER] saw strength in [segment], does that read through to peers?
- Any supply chain or demand signals that affect the group?
- How should I adjust my expectations for peers reporting next week?
This is the read-through that makes earnings season a compounding advantage instead of a survival exercise. Every call you cover feeds context into the next one.
Why This Works
Earnings prep is not hard — it is repetitive. The analysis itself is pattern recognition: compare this quarter's numbers to prior quarters, check guidance vs history, flag anomalies. LLMs are excellent at exactly this.
The key insight: you are not asking the AI to make investment decisions. You are asking it to organize information so YOU can make better decisions faster. That is the difference between "AI trading" and actually using AI in finance.
Real Numbers
Before this workflow: ~20 minutes of prep per name during earnings season. With 40 names reporting over 2 weeks, that is 13+ hours of prep.
After this workflow: ~90 seconds per name for the brief, plus 5 minutes of review for my highest-conviction names. Total time: under 2 hours for the same coverage.
That is 11 hours back. Every quarter.
What is Coming Next Week
Issue #2: The portfolio risk scan that runs every morning before the market opens. One prompt, 60 seconds, catches the thing you would have missed at 4 PM.
AI Finance Brief is written by a team that runs live algorithmic trading systems daily. Every workflow is tested on real market data before it reaches your inbox.
Free every week. Subscribe at aifinancebrief.com